pkn Posted April 23, 2015 Share Posted April 23, 2015 They're in. I've blown away my old array and started again with the two 8TB drives. Parity is building at 189MB/s, expected to finish in just under 12 hours. 15 hours more like it. It's 190 MB/s on outer cylinders, speed will fall to ~90 MB/s on inner cylinders. Quote Link to comment
HellDiverUK Posted April 23, 2015 Share Posted April 23, 2015 34% done, 4 hours 12 minutes done, 8 hours 4 mins to go. 179.9MB/s at the moment. Quote Link to comment
danioj Posted April 24, 2015 Share Posted April 24, 2015 Thanks for the update ... just wondered what had happened since the status posts suddenly went "silent". Just wanted to be sure you didn't have any system crashes; failed validations; etc. Bad news is NOT allowed :) ... these Archive drives are doing really well !! Hi Guys, back home in Rainy Melbourne Anyway, just started the copy of the "smaller files" going. Another ~4TB made up of 22,526 Files. Started at 0% as always 40MB/s. Will provide periodic updates. Quote Link to comment
garycase Posted April 24, 2015 Share Posted April 24, 2015 They're in. I've blown away my old array and started again with the two 8TB drives. Parity is building at 189MB/s, expected to finish in just under 12 hours. Are these the only two drives in your new array? Correct. For now, anyway. The two 4TB drives will go in later. Note that this will skew your performance measurements, as writes will be much faster than they would be if the array had at least 2 data drives. UnRAID recognizes the special case of a 2-drive array, and then does writes the same as a standard RAID-1 setup => i.e. a write will require 2 disk operations instead of the 4 operations normally required. Quote Link to comment
garycase Posted April 24, 2015 Share Posted April 24, 2015 Hi Guys, back home in Rainy Melbourne Anyway, just started the copy of the "smaller files" going. Another ~4TB made up of 22,526 Files. ... Will provide periodic updates. Welcome back. I anticipate the performance will be just as good with these as it was with the larger files; but it will be nice to confirm this. Quote Link to comment
HellDiverUK Posted April 24, 2015 Share Posted April 24, 2015 Parity build ended in a little under 12 hours 30 minutes. Copying stuff over at the moment at 100MB/s more or less... Quote Link to comment
garycase Posted April 24, 2015 Share Posted April 24, 2015 Parity build ended in a little under 12 hours 30 minutes. Copying stuff over at the moment at 100MB/s more or less... Nice speed -- continues to confirm what Daniel's tests have already been showing ... these drives do VERY well as long as you're copying a lot of sequential data. Seagate did a very nice job of mitigating the shingled architecture's failings. Note that your write speeds will be MUCH lower when you add more data drives; as writes will then be done in the "normal" way, which requires 4 disk operations per write, whereas right now you're writing to what's effectively a RAID-1 array, so there are 2 disk operations per write and they're both writes, so there's no intervening latency delay for an extra rotation of the platter. Quote Link to comment
pkn Posted April 24, 2015 Share Posted April 24, 2015 Parity build ended in a little under 12 hours 30 minutes. ... I'm jealous! Quote Link to comment
garycase Posted April 24, 2015 Share Posted April 24, 2015 Parity build ended in a little under 12 hours 30 minutes. Copying stuff over at the moment at 100MB/s more or less... Good speed. Daniel's took about 15:10 ... but he had 3 of the disks -- not just 2. I suspect this is another case where the special case of just 2 drives helped, since it effectively just wrote a RAID-1 initialization instead of actually computing parity. Quote Link to comment
pkn Posted April 24, 2015 Share Posted April 24, 2015 Parity build ended in a little under 12 hours 30 minutes. Copying stuff over at the moment at 100MB/s more or less... Good speed. Daniel's took about 15:10 ... but he had 3 of the disks -- not just 2. I suspect this is another case where the special case of just 2 drives helped, since it effectively just wrote a RAID-1 initialization instead of actually computing parity. I don't think computing per se would make any noticeable difference, it's so much faster than any IO operation. My bet would be on drives rotating not completely in sync (unRAID is not RAID, so it does not force synchronized rotation, doesn't it?) so mdrecoveryd have to wait for some of the drives extra turn from time to time. If I'm right, then parity sync with 4 drives (parity+3 data) would take even more than 15+ hours... hmm... I'd have to buy a third license to test this... or temporarily borrow licensed flash from one of the production servers. Quote Link to comment
garycase Posted April 24, 2015 Share Posted April 24, 2015 The difference in the special 2-disk case is that the drives are treated like a RAID-1 array => so there's no need to compute the parity. This means it can simply copy the data disk to the parity disk -- no computations required. With more than 2 drives, it first has to read all of the drives except the parity drive; compute the parity bits; and then write the parity drive. The computational overhead isn't large, but it's enough that the parity drive will likely have some rotational latency before it starts writing the cylinder. Once you have 2 data drives, adding more doesn't make much difference, since all the reads are done at the same time. Slight rotational offsets don't really matter, as the drives that finish early can easily "catch up" with the interface-rate transfers from their buffers once the parity calculations are started. Computational overhead may make a slight difference with larger arrays ... but not a lot. For example, Daniel's parity sync took 15:10 with 3 drives. I'd expect it to still be 15:xx no matter how many drives were in the array ... the xx would probably be slightly more than 10, but not by a lot. Quote Link to comment
danioj Posted April 24, 2015 Share Posted April 24, 2015 Thanks for the update ... just wondered what had happened since the status posts suddenly went "silent". Just wanted to be sure you didn't have any system crashes; failed validations; etc. Bad news is NOT allowed :) ... these Archive drives are doing really well !! Hi Guys, back home in Rainy Melbourne Anyway, just started the copy of the "smaller files" going. Another ~4TB made up of 22,526 Files. Started at 0% as always 40MB/s. Will provide periodic updates. Update: 41% into the copy: 1.5TB down of ~4TB. (10,422 files of 22,526) so far. Still speeding along at ~40MB/s. Quote Link to comment
danioj Posted April 25, 2015 Share Posted April 25, 2015 Update: 41% into the copy: 1.5TB down of ~4TB. (10,422 files of 22,526) so far. Still speeding along at ~40MB/s. Update: 60% into the copy: 2.3TB down of ~4TB. (13,251 files of 22,526) so far. Still speeding along at ~40MB/s. Quote Link to comment
garycase Posted April 25, 2015 Share Posted April 25, 2015 Thanks for the updates .. definitely looking very good. Quote Link to comment
danioj Posted April 25, 2015 Share Posted April 25, 2015 Update: 60% into the copy: 2.3TB down of ~4TB. (13,251 files of 22,526) so far. Still speeding along at ~40MB/s. Update: 80% into the copy: 3.05TB down of ~4TB. (19,000 files of 22,526) so far. Still speeding along at ~40MB/s. In addition - I monitored the copy in the test of the ~4GB files and have done so in this test too. INTERESTING results! Where I saw periodic drops in speed as it reached 100% of 4TB of the larger files I am now seeing INCREASES in speed as it is reaching 4TB of the smaller files up to ~59MB/s. I am certain these speed fluctuations are not material as they do not impact the mean speed of the copy. But I find them interesting all the same. For those who need more evidence: P.S. I am 20% progress (to completion) away from concluding that these drives are FINE for my use in an UNRAUD environment. I have thousands of files all on average between 400MB and 15GB and between two copies of 4TB of sets of both I have experienced NO difference in write or read speed between these Seagate 8TB drives and my existing WD Red 3TB drives. P.P.S. The above also means I am 20% progress (to completion) away from using these drives in the refresh of my Main Server over my planned 6TB WD drives. Quote Link to comment
pras1011 Posted April 25, 2015 Share Posted April 25, 2015 Have you tried a huge file yet? 20gb+? Quote Link to comment
danioj Posted April 25, 2015 Share Posted April 25, 2015 Have you tried a huge file yet? 20gb+? My largest file is a 40,275,436 KB (40GB) file which was transferred in the last test I did (and which I monitored) and it transferred in FULL at ~40MB/s. Quote Link to comment
pras1011 Posted April 25, 2015 Share Posted April 25, 2015 Oh yes thanks. This is great news. How much power do they consume spun down, spun up, read/write etc? And aren't you worried about the reliability of Seagate? Quote Link to comment
danioj Posted April 25, 2015 Share Posted April 25, 2015 Oh yes thanks. This is great news. How much power do they consume spun down, spun up, read/write etc? And aren't you worried about the reliability of Seagate? I haven't done any tests on power consumption - BUT - I am happy to do them if someone shows me how they would like the test to be run and how they want the data to be presented! As for reliability of Seagate (and please ALL remember this is my opinion based on my observations ONLY over the years in an Enterprise and Personal environment) - I don't really subscribe to the whole "Seagate drives always fail and are generally less reliable". Generally my observations have led me to conclude that in the main those who have had drives fail (Seagate or otherwise) inside the warranty of the drive tend to be those who don't adequately "work out" their drives before deploying them OR operate them outside the manufacturers operating conditions (e.g. temperature)! Even I didn't work out my past drives like I've given these bad boys a work out (This is down to me learning some best practices off of the likes of Gaz and Brian). I know that doesn't say anything to their future reliability (albeit these newly learned methods for ensuring a drive passes the infant mortality period have been applied) BUT then again they come with a 3 year warranty. On top of Unraid's fault tollerance I have a COMPLETE backup of my data. ALL Drives WILL Fail, it's just a matter of time - if they do Ill replace them - if they are in Warranty Ill RMA if not I've had my moneys worth! My Data is safe so I am good. I'm quite happy. Quote Link to comment
pras1011 Posted April 25, 2015 Share Posted April 25, 2015 How many preclears did you run on each drive? I am planning to replace my drives (in signature) with 8 of these. My Hitachi are extremely reliable but they should be for enterprise drives. But there are too many of them and my server case is huge. Quote Link to comment
danioj Posted April 25, 2015 Share Posted April 25, 2015 How many preclears did you run on each drive? 3.5 Pre Clears (I had a power cut in the middle of the first run) and a LONG S.M.A.R.T test at the end - which incidentally is now my default pre deployment drive checking regime. On top of that of course a Parity Sync and then a Parity Check when I configured the Array before any data was copied to the Array. All Good. Quote Link to comment
pras1011 Posted April 25, 2015 Share Posted April 25, 2015 How long does it take to do one preclear? Quote Link to comment
danioj Posted April 25, 2015 Share Posted April 25, 2015 How long does it take to do one preclear? Link to post earlier in the thread: http://lime-technology.com/forum/index.php?topic=36749.msg363800#msg363800 ~ 52 Hours (Using Brian's FASTER preclear script). Quote Link to comment
pras1011 Posted April 25, 2015 Share Posted April 25, 2015 Oh nuts! 156 hours per hdd!! And therefore use a few sessions to preclear several hdds together. Quote Link to comment
danioj Posted April 25, 2015 Share Posted April 25, 2015 Oh nuts! 156 hours per hdd!! Yep. Note: that I run the preclears for all 3 drives in Parallel within a separate tty using the bmi interface provided by the ASRock board (benefits of Server grade H/W ). But you could achieve the same using SCREEN. Quote Link to comment
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.